State lawmakers are moving aggressively to regulate artificial intelligence in health care, filling a vacuum left by the absence of AI-specific federal rules, according to an American Medical Association analysis. Legislatures introduced more than 100 health-care AI bills in 2024 and passed 20 of them, with momentum expected to continue through 2025. California now requires physicians and health systems to disclose when generative AI is used in patient communications, mandates human review before health plans deny coverage, and imposes broader transparency duties on large model developers. Colorado adopted sweeping consumer protections that place significant responsibilities on developers and deployers of high-risk AI tools, while Utah enacted narrower measures that nonetheless compel disclosure when generative AI is used by regulated occupations, including clinicians. Many of the new laws target automated decision-making by insurers and hospitals, instructing that any AI tool recommending a limitation or denial of care be reviewed by a licensed physician in the relevant specialty. The AMA, which recently updated its policy on AI, warns that unchecked automation can increase denial rates for medically necessary treatments and erect barriers to timely care. With rapid advances in generative models outpacing traditional regulatory processes, the physicians’ group is urging states to balance innovation with stronger patient safeguards. It says health-care organizations should establish multidisciplinary AI governance committees, require rigorous vendor due diligence, and maintain the ability for clinicians to override algorithmic recommendations.